Multilayer universal video coder

Abstract

A video coder is disclosed which codes the digital pel values of a video signal in such a manner that different levels of picture quality, i.e. resolution, are available to users. The coder uses a two-dimensional quadrature-mirror filter (103) to decomposes the input into plural subbands. The baseband lowest horizontal-lowest vertical frequency subband is coded using a hybrid DCT/DPCM coder (104) that is compatible with a proposed CCITT standardized codec. The output of this coder is a layer 1 signal which is available to subscribers who desire only low-quality video service. The other subbands are coded using an interframe DPCM coder (108, 109, 110, 111) for those subbands in which the filtered samples in the subband represent horizontal or vertical edge variations in the video frame, or an intraframe PCM coder (116, 117) for those subbands in which the filtered samples in the subband represent diagonal variations. Other layer signals are formed by combining the outputs of selected of the interframe and intraframe codes. A user subscribing to increasing number of signal layers is able to reconstruct a signal of increasing quality.

What is claimed is: 
     
       1. A signal coder for coding a video signal into plural layer output signals, the signal coder comprising: means for receiving an input signal consisting of digital pel values of a scanned video signal, said video signal consisting of sequential video frames, each video frame consisting of a plurality of scan lines in the horizontal direction, each scan line consisting of a plurality of pels, the pels in each frame forming rows in the horizontal direction and columns in a vertical direction perpendicular to the horizontal direction;   quadrature-mirror filter means for decomposing in the horizontal direction along each row and in the vertical direction down each column the pel values of each current video frame of pel values into plural narrow band subband sequences of filtered pel values;   first interframe coding means for coding the lowest horizontal-lowest vertical frequency subband of filtered pel values of the current frame using the lowest horizontal-lowest vertical frequency subband of filtered pel values of a previous frame to form a first layer output signal;   plural secondary coding means each one of coding one of the plural narrow band subband sequences of filtered pel values of the current frame other than the subband of filtered pel values coded by said first interframe coding means, each one of said secondary coding means being either an intraframe-type coder which directly codes the filtered pel values of the current frame or an interframe-type coder which codes differences between the filtered pel values of the current frame and corresponding filtered pel values of a previous frame, and;   at least one combining means for combining the coded outputs of selected of said plural secondary coding means to form at least a second layer output signal;   wherein video signal of increasingly higher spatial resolutions can be reconstructed from combinations of increasing numbers of the layer output signals.   
     
     
       2. A signal coder in accordance with claim 1 wherein said first interframe coding means for coding the lowest-horizontal lowest vertical frequency subband is hybrid discrete cosine transform (DCT)/differential pulse code modulation (DPCM) coding means. 
     
     
       3. A signal coder in accordance with claim 2 wherein each of the subband sequences of filtered pel values of those bands that correspond to the horizontal and vertical edge variations in the input video frame, are coded by one of said plural secondary coding means that is an interframe-type coder, and each of the subband sequences of filtered pel values of those bands that correspond to the diagonal variations in the input video frame are coded by one of said plural secondary coding means that is an intraframe-type coder. 
     
     
       4. A signal coder in accordance with claim 3 further comprising overhead information processing means for deriving motion displacement information from said hybrid DCT/DPCM coding means and for supplying the derived motion displacement information to each one of said plural secondary coding means that is an interframe-type coder. 
     
     
       5. A signal coder in accordance with claim 2 wherein said quadrature-mirror filter means nonuniformly decomposes in the horizontal and vertical directions that pel values of each video frame of pel values into seven subbands of filtered pel values comprising a low horizontal-low vertical-low horizontal-low vertical (LLLL) frequency subband, a low horizontal-low vertical low-horizontal high-vertical (LLLH) frequency subband, a low-horizontal low-vertical high-horizontal low-vertical subband (LLHL) frequency subband, a low-horizontal low-vertical high-horizontal high-vertical (LLHH) frequency subband, a low-horizontal high vertical (LH) subband, a high-horizontal low-vertical (HL) frequency subband, and a high-horizontal high-vertical (HH) frequency subband, said hybrid DCT/DPCM coding means coding said LLLL frequency subband, and one each of said plural secondary coding means that is an interframe-type coder coding each of said LLLH, said LLHL, said LH and said HL fluency subbands, and one each of said plural secondary coding means that is an intraframe-type coder coding each of said LLHH and said HH frequency subbands, a first of said at least one combining means for combining the coded outputs of said LLLH, said LLHL, and said LLHH frequency bands to form a second layer output signal, and a second of said at least one combining means for combining the coded outputs of said LH, said HL, and said HH frequency bands to form a third layer 
     
     
       6. A method of coding a video signal into plural layer output signals comprising the steps of: receiving an input signal consisting of digital pel values of a scanned video signal consisting of sequential video frames, each video frame consisting of a plurality of scan lines in the horizontal direction, each scan line consisting of a plurality of pels, the pels in each frame forming rows in the horizontal direction and columns in a vertical direction perpendicular to the horizontal direction;   decomposing in the horizontal direction along each row and in the vertical direction down each column the pel values of each current video frame of pel values into plural narrow band subband sequences of filtered pel values;   coding the filtered pel values in the lowest horizontal-lowest vertical frequency subband of the current frame using the lowest horizontal-lowest vertical frequency subband of filtered pel values of a previous frame to form a first of a plurality layer output signal;   coding each of the plural narrow band subband sequences of filtered pel values of the current frame other than the lowest horizontal-lowest vertical frequency subband using either an interframe-type coder or an intraframe-type coder;   combining selected of the coded narrow band subband sequences of pel values of the current frame other then the lowest horizontal-lowest vertical frequency subband to form at least a second layer output signal;   wherein video signals of increasingly higher spatial resolution can be reconstructed from increasing numbers of the layer output signals.   
     
     
       7. The method of claim 6 wherein a hybrid DCT/DPCM interframe coder is used to code the filtered pel values in the lowest horizontal-lowest vertical frequency subband of the current frame.